PREDICTION OF TRADING PROFIT OF TRANSNATIONAL COMPANY USING ARTIFICIAL NEURAL NETWORKS: A CASE STUDY OF NESTLE IN EUROPE
SVITLANA GALESHCHUK *
Ternopil National Economic University, 11 Lvivska Str., Ternopil, 46000, Ukraine.
*Author to whom correspondence should be addressed.
Abstract
Application of artificial neural networks for economic forecasting is described and empirically examined with Nestle financial reporting data. For the experiments, panel data of the exchange rates as well as trading profit, volume of sales, currency retranslations, and effects of exchange rate changes are used to predict future quarterly profits with neural networks. The best neural network model is found with the best forecasting abilities, based on a mean absolute percentage error measure. Values of prediction errors (mean and maximum errors for 8 quarters of 2013-2014 do not exceed 5.4% and 8.7% respectively) show that artificial neural networks can provide accurate prediction results for international firms’ profit. Accurate prediction results can provide improved strategic management of international companies that conduct operations in foreign currencies.
Keywords: Neural networks, multilayer perceptron, profit prediction, currency risk